AI-ordlista
Den kompletta ordlistan över AI
Model Predictive Control (MPC)
Advanced control strategy that uses a dynamic model of the HVAC system to anticipate future conditions and optimize energy setpoints over a sliding time horizon.
Reinforcement Learning for HVAC
Application of AI algorithms where an agent learns an optimal control policy (e.g., valve adjustment, fan speed) through trial and error to maximize a defined reward (e.g., comfort - energy cost).
Building Digital Twin
Dynamic virtual replica of a building and its HVAC systems, fed in real-time by sensors, used to simulate, predict, and optimize energy performance before applying actions in the real world.
Self-adaptive Regulation
Control system that automatically adjusts its internal parameters (e.g., PID constants) based on environmental variations (weather, occupancy) to maintain optimal performance without manual intervention.
Thermal Load Forecasting
Use of machine learning models to anticipate a building's heating or cooling demand over a short-term horizon, based on weather forecasts, calendar, and occupancy history.
Building-to-Vehicle (B2V) Integrated Energy Management
Control strategy that coordinates building HVAC systems with electric vehicle charging/discharging to optimize local energy self-consumption and reduce peak power drawn from the grid.
Energy Cost Function
Mathematical expression quantifying the energy expenditure of an HVAC system, often weighted by dynamic electricity tariffs, that optimization algorithms seek to minimize.
Hygrothermal Model
Simulation model that couples heat and moisture transfers in a building envelope, essential for optimizing HVAC systems that handle both temperature and humidity.
Stochastic Gradient Descent for HVAC
Iterative optimization algorithm used to adjust the parameters of a predictive control model using random subsets of data, accelerating convergence for complex systems.
Fuzzy Logic for Comfort
Control method that translates subjective comfort variables (e.g., 'a bit warm', 'too humid') into precise control actions on HVAC equipment, mimicking human reasoning.
Recurrent Neural Networks (RNN) for Energy Time Series
Deep learning architecture specialized in predicting sequential data, such as the power consumption of an HVAC system, by memorizing long-term temporal dependencies.
Online vs Offline Optimization
Distinction between optimization strategies calculated periodically on historical data (offline) and those adjusted in real-time in response to new measurements (online) for HVAC control.
Thermal Load Shedding Strategy
AI-controlled action plan that preemptively and temporarily reduces the heating or cooling power to smooth the load curve and reduce costs during peak hours.
Prediction Horizon
Future time period over which a predictive control algorithm (MPC) performs its optimizations, a crucial trade-off between prediction accuracy and potential for HVAC energy savings.
State Space Mapping
AI technique that explores and models the set of all possible states of an HVAC system (temperature, pressure, flow rate) to identify optimal trajectories towards a target state of comfort and efficiency.